Multiple connection methods
Connect AO2 through MCP, API key, SDK, or CLI so different AI tools and apps can read and write to the same memory layer.
AO2 Memory is a shared memory layer for AI tools, reusing context across agents via MCP, API key, SDK, or CLI.
AO2 Memory is a shared memory layer for AI tools. The product is positioned as a way to avoid re-explaining the same project context every time you switch between agents, so one tool can write memory that another can read later.
The site describes AO2 as an MCP server you connect to tools such as Claude, ChatGPT, Codex, OpenClaw, Hermes, Manus, or your own apps through an API. It emphasizes structured memory that you control and can export, rather than a hidden store locked inside a single assistant.
Connect AO2 through MCP, API key, SDK, or CLI so different AI tools and apps can read and write to the same memory layer.
Use the same memory across Claude, ChatGPT, Codex, OpenClaw, Hermes, Manus, and custom apps connected by API.
See changes made by agents, review activity, and roll back earlier updates when needed.
Scope access to specific records or fields, and issue temporary keys that expire automatically.
Store information as structured records instead of an unsearchable text blob, using fields such as title, status, due date, priority, and number.
Export memory with one command and move it out as Git, JSON, Markdown, or CSV.
Keep project context in one place so switching from one AI assistant to another does not require repeating the same background information.
Track agent edits over time, inspect what changed, and undo updates when an AI writes something you do not want to keep.
Store notes and project data as records with fields instead of dumping them into plain text, making the memory easier to reuse later.
Connect custom apps or internal workflows through the API when you want AI memory available outside a single chat interface.
Export memory to Git or common file formats when you want to keep ownership of your context outside the product.
AO2 Memory is set up by connecting it to an AI tool through MCP, or by using an API key. The page also mentions SDK and CLI options for connecting it to apps and workflows.
The site says AO2 Memory works with Claude, ChatGPT, Codex, OpenClaw, Hermes, Manus, and custom apps via API. It is presented as a shared memory layer across those tools.
The pricing page says plans are based on active records across the account, and archived records stay available without counting toward the limit. The site does not provide setup details for teams or collaboration beyond that.
AO2 Memory stores structured records rather than a free-form text dump. The page shows fields such as title, text, status, due date, priority, and number as examples of the kind of structure it can support.